CN117039840A - Feature automatic enhancement-based household load identification method and system - Google Patents
Feature automatic enhancement-based household load identification method and system Download PDFInfo
- Publication number
- CN117039840A CN117039840A CN202310759240.9A CN202310759240A CN117039840A CN 117039840 A CN117039840 A CN 117039840A CN 202310759240 A CN202310759240 A CN 202310759240A CN 117039840 A CN117039840 A CN 117039840A
- Authority
- CN
- China
- Prior art keywords
- load
- probability
- model
- current
- characteristic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000005611 electricity Effects 0.000 claims abstract description 45
- 238000004364 calculation method Methods 0.000 claims abstract description 30
- 238000005070 sampling Methods 0.000 claims description 35
- 239000000203 mixture Substances 0.000 claims description 21
- 238000012937 correction Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 17
- 230000006870 function Effects 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 9
- 238000012795 verification Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 5
- 230000002708 enhancing effect Effects 0.000 claims description 2
- 238000012512 characterization method Methods 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 235000013421 Kaempferia galanga Nutrition 0.000 description 1
- 244000062241 Kaempferia galanga Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005684 electric field Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/003—Load forecast, e.g. methods or systems for forecasting future load demand
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Power Engineering (AREA)
- Probability & Statistics with Applications (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application belongs to the technical field of electric loads, and particularly relates to a household load identification method and system based on characteristic automatic enhancement. The application provides a load identification method and system based on characteristic automatic enhancement, which are characterized in that electric and operation information of an electric load is collected and recorded, the electric and operation information is input into a load characteristic generation module, VI tracks and expert characteristics are generated, a load characteristic data set is formed and divided, a characteristic enhancement and identification model is trained, the trained model is utilized for carrying out load identification, a preliminary identification result is obtained, the electric probability model is generated by utilizing the load operation information, the load identification result is corrected, and a final load identification result is obtained. Aiming at the problems of low recognition precision and poor method robustness in the current household load recognition field, the application effectively enhances the characterization capability of the load characteristics on the premise of smaller calculation amount by designing the characteristic self-adaptive enhancement network, thereby improving the accuracy of the load in a complex electricity utilization environment.
Description
Technical Field
The application belongs to the technical field of electric loads, and particularly relates to a household load identification method and system based on characteristic automatic enhancement.
Background
The realization of efficient and safe electricity utilization at the electricity utilization side becomes an important subject for the development of the power grid. The load identification technology is used as an important part in the intelligent electricity consumption sensing technology, and can obtain detailed electricity consumption information of resident users by excavating and analyzing the user electrical signals collected by the intelligent terminal, so that the load identification technology has important effects on aspects of user side participation demand response, family energy management, dangerous electricity consumption early warning and the like. In terms of demand response, the load identification technology is helpful for identifying the type of electric appliances and the electricity consumption demand in the power grid, so that real-time monitoring and optimization of the power system are realized. Through accurate grasp of electricity demand, the electric power system can realize more efficient energy scheduling and distribution, reduce carbon emission and push energy transformation; in the aspect of household energy management, the load identification technology can help a user to know the electricity consumption condition and the electricity consumption efficiency of household electric equipment, so that the user is guided to implement energy-saving measures, and the household energy utilization rate is improved. The household energy consumption is optimized, the energy cost is reduced, and the life quality is improved; in the aspect of dangerous electricity utilization early warning, the load identification technology can monitor the running condition of an electric appliance in real time and timely discover abnormal electricity utilization phenomenon, so that potential safety hazards are prevented. The method is not only helpful for guaranteeing the life and property safety of people, but also can avoid the waste of electric power resources and environmental pollution caused by electricity utilization accidents.
In the field of load recognition, there are many methods for recognizing an electric signal by imaging, and a good effect is obtained. Three-channel color VI track images are constructed as in the literature 'non-invasive load identification method based on V-I track color coding', and load identification is realized by using an Alexnet network; the document Temporal and Spectral Feature LearningWith Two-Stream Convolutional Neural Networks for Appliance Recognition in NILM adopts a method for improving the Galangal angle field to image the single load waveform from the time domain and the frequency domain respectively, and uses a double-flow neural network to carry out load identification; document Adaptive Weighted Recurrence Graphs forAppliance Recognition in Non-Intrusive Load Monitoring uses load data to construct an adaptive weighted recursion map for identification. However, in general, the current load identification method still has the problems of complex identification precision and generalization, complex identification model, low practicability and the like. Therefore, the method of feature enhancement can be considered to adaptively enhance the extracted load features, so that the recognition network focuses on the important part of the features, and the recognition accuracy of the load can be further improved.
Based on the above, the patent provides a load recognition system and a method based on feature automatic enhancement, which are characterized in that an encoder network with the feature automatic enhancement is designed, an adaptive feature enhancement matrix is generated by using expert electrical features, the feature enhancement is carried out on VI track images, and the recognition performance of a model is improved. Meanwhile, a classification correction module is designed according to the load operation information, and the identification result is verified according to the operation rule and the use characteristics of the load, so that the judgment logic of the load identification is further improved.
Disclosure of Invention
This section is intended to outline some aspects of embodiments of the application and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section as well as in the description of the application and in the title of the application, which may not be used to limit the scope of the application.
The present application has been made in view of the above-described problems.
Therefore, the technical problems solved by the application are as follows: the household load identification method based on the characteristic automatic enhancement is provided, the characteristic of the load is enhanced by designing a characteristic self-adaptive enhancement network, and a classification correction module is constructed to identify and correct the load with the operation rule contrary to the identification type, so that the problems of low characterization capability and poor identification accuracy of the load type in a complex electricity utilization environment are solved.
In order to solve the technical problems, the application provides the following technical scheme: a feature-based automatic enhancement home load identification method, comprising: collecting operation information of the load, and recording the collected operation information; inputting the collected operation information into a load characteristic generation module, generating VI track image characteristics and expert electrical characteristics, and forming and dividing an electricity load characteristic data set; and classifying and correcting the collected operation information to generate an electricity probability model, and carrying out probability calculation to obtain a final load identification result.
As a preferred embodiment of the feature-based automatic enhancement home load recognition method of the present application, wherein: the operation information comprises voltage and current signals of a known load, and records the time period of signal acquisition, the position of a sampling terminal and the operation information of the duration of load operation;
the voltage and current signals of the load comprise load steady-state voltage and current data with sampling frequency larger than 6.4kHz and sampling period larger than 10 power frequency periods.
As a preferred embodiment of the feature-based automatic enhancement home load recognition method of the present application, wherein: the VI track image features comprise image features formed by load voltage and current tracks, and the construction method comprises the following steps:
normalizing the load voltage and current waveforms:
wherein U is max 、I max Is the maximum value of voltage and current in one cycle, U min 、I min Is the minimum value of voltage and current in one cycle;
constructing a k x k zero matrix M with resolution k, and normalizing the voltageCurrent->And multiplying the resolution k, rounding downwards, taking the current value as a row, taking the voltage value as a column, and filling the corresponding position of the matrix M.
As a preferred embodiment of the feature-based automatic enhancement home load recognition method of the present application, wherein: the expert electrical characteristics comprise manual characteristics calculated according to load voltage and current data;
the manual characteristics comprise current peak value, current effective value, active power value, reactive power value, current harmonic distortion rate, current waveform coefficient, amplitude and phase of 1 st, 3 rd and 5 th current harmonic;
the current harmonic information is calculated by FFT, and the current peak value I is pp Calculation, expressed as:
I pp ＝max{i}-min{i}
max { i } is the maximum value in the current cycle of the sample, and min { i } is the minimum value in the current cycle of the sample;
effective value of current I rms Calculation, expressed as:
wherein i is n An instantaneous value of an nth sampling point in the current waveform;
active power P is calculated, expressed as:
wherein v is n An instantaneous value of an nth sampling point in the voltage waveform;
active power S is calculated, expressed as:
reactive power Q calculation, expressed as:
wherein u is n An instantaneous value of an nth sampling point in the voltage waveform;
power factor PF calculation, expressed as:
current harmonic distortion rate I THD Calculation, expressed as:
current waveform coefficient I wave Calculation, expressed as:
wherein N is the sampling frequency in one power frequency period.
As a preferred embodiment of the feature-based automatic enhancement home load recognition method of the present application, wherein: dividing the power consumption load characteristic data set comprises dividing the generated power consumption load characteristic data set into a training set and a verification set, automatically enhancing and identifying module parameters by training and adjusting characteristics, and stopping model training when the accuracy of the model in the verification set reaches a threshold value;
the feature enhancement network and the load identification network adopt an integral training mode, and a cross entropy loss function is adopted as a loss function of the feature enhancement network and the load identification network to train a model, and the model is expressed as follows:
wherein L is the total number of samples, M is the number of sample categories, y ic As a sign function, 1 is taken if the predicted class of the sample i is the same as the true class, 0 is taken if the predicted class of the sample i is different from the true class, and p is ic The probability that sample i belongs to category c.
As a preferred embodiment of the feature-based automatic enhancement home load recognition method of the present application, wherein: the electricity probability model comprises the following specific construction steps,
classifying known loads according to types, dividing a day into 24 time periods according to hours, counting the power consumption probability of each type of load in different time periods, and simulating a power consumption time period probability curve by using a Gaussian mixture model to obtain a power consumption time period model of the type c load, wherein the power consumption time period model is expressed as follows:
wherein n is GaussianNumber of models, parameter θ 1 Comprises mean mu 1 ,…,μ n And standard deviation sigma 1 ,…,σ n ；
Fitting an electricity consumption time length probability curve of each type of load through a Gaussian mixture model, and representing the electricity consumption time length probability model of the type c load as phi 2c (x|θ 2 )；
For a home environment, the user potential is divided into: the living room, bedroom, dining room and bathroom 3 parts take the frequency of various types of loads in each environment as probability to obtain probability models of various types of loads in different positions, and the potential probability model of the type c load is expressed as phi 3c (x|θ 3 )；
For the load x identified as the type c, according to the collected electricity consumption period, electricity consumption duration and electricity consumption position, calculating to obtain the probability that the load x belongs to the type c, wherein the probability is expressed as follows:
P(x∈c)＝φ 1c (x|θ 1 )·φ 2c (x|θ 2 )·φ 3c (x|θ 3 )
wherein phi is 1c Gaussian probability model, θ, of the period of power consumption for type c loads 1 Is phi 1c Model parameters phi of (2) 2c Gaussian probability model, θ, of the power duration mixture for type c loads 2 Is phi 2c Model parameters phi of (2) 3c Potential probability model for type c load, θ 3 Is phi 3c Is used for the model parameters of the model.
Normalizing the probability of load generation to be distinguished and the probability of load generation of the known same type together to obtain the probability of each type output by the final classification correction module, wherein the probability is expressed as follows:
wherein P' represents the normalized probability value of the load to be discriminated, P represents the original probability value of the load to be discriminated, P max Representing the maximum of all the same type of load generation probabilities.
As described in the applicationA preferred embodiment of the feature-based automatic enhancement home load recognition method, wherein: the Gaussian mixture model comprises a maximum expected algorithm as a fitting method, and the number k of the Gaussian mixture model is determined by setting a selection interval [ k ] of k min ,k max ]Sequentially select k min To k max Fitting is carried out on the number of Gaussian models, the number of Gaussian models when the correlation coefficient r of the fitting result is maximum is selected as the number k of the selected Gaussian models, and a calculation formula of the correlation coefficient r is expressed as follows:
wherein n is the length of the sequence, x is the actual load information data distribution, y is the fit probability curve of the Gaussian mixture model,for the mean value of the actual load information data distribution, +.>And fitting the mean value of the probability curve to the Gaussian mixture model.
Another object of the present application is to provide a feature-based automatic enhancement home load identification system, which can enhance load features by designing a feature-adaptive enhancement network, and construct a classification correction module to identify and correct loads with operation rules contrary to identification types, so as to solve the problems of low characterization capability and poor identification accuracy of load types in complex electricity utilization environments.
In order to solve the technical problems, the application provides the following technical scheme: a household load identification system based on characteristic automatic enhancement comprises a load information acquisition module, a load characteristic generation module, a characteristic automatic enhancement and identification module and a classification correction module;
the load information acquisition module is used for acquiring the running information of household loads;
the load characteristic generation module is used for generating VI track image characteristics and expert electrical characteristics to form an electricity load characteristic data set;
the characteristic automatic enhancement and recognition module is used for fusing the manually extracted electric characteristics and VI track image characteristics to obtain characteristics subjected to characteristic enhancement, and inputting the characteristics into the classification network to realize automatic recognition of different load types;
the classification correction module establishes power consumption probability models of different types of loads by adopting signal acquisition time period information, running position information and running time length information of the existing loads, and further carries out probability calculation on the judging results of the automatic characteristic enhancement and the identification module according to the information of the loads to be identified.
A computer device comprising a memory and a processor, said memory storing a computer program, characterized in that said processor, when executing said computer program, implements the steps of a feature-based automatic enhanced home load identification method.
A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of a feature-based automatic enhanced home load identification method.
The application has the beneficial effects that: the application provides a load identification method and system based on characteristic automatic enhancement, which aim at the problems of low identification precision and poor method robustness in the existing household load identification field, enhance the load characteristics by designing a characteristic self-adaptive enhancement network, and construct a classification correction module to identify and correct the load with the operation rule contrary to the identification type.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:
FIG. 1 is a flow chart of a feature-based automatic enhancement home load identification method according to one embodiment of the present application;
FIG. 2 is a block diagram of a feature enhanced network and a load identification network of a feature-based automatic enhanced home load identification method according to an embodiment of the present application;
fig. 3 is an overall structure diagram of a home load recognition system based on feature automatic enhancement according to an embodiment of the present application.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present application can be understood in detail, a more particular description of the application, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application, but the present application may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present application is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the application. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
While the embodiments of the present application have been illustrated and described in detail in the drawings, the cross-sectional view of the device structure is not to scale in the general sense for ease of illustration, and the drawings are merely exemplary and should not be construed as limiting the scope of the application. In addition, the three-dimensional dimensions of length, width and depth should be included in actual fabrication.
Also in the description of the present application, it should be noted that the orientation or positional relationship indicated by the terms "upper, lower, inner and outer", etc. are based on the orientation or positional relationship shown in the drawings, are merely for convenience of describing the present application and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present application. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The terms "mounted, connected, and coupled" should be construed broadly in this disclosure unless otherwise specifically indicated and defined, such as: can be fixed connection, detachable connection or integral connection; it may also be a mechanical connection, an electrical connection, or a direct connection, or may be indirectly connected through an intermediate medium, or may be a communication between two elements. The specific meaning of the above terms in the present application will be understood in specific cases by those of ordinary skill in the art.
Example 1
Referring to fig. 1 and 2, for one embodiment of the present application, there is provided a home load recognition method based on feature automatic enhancement, including:
collecting operation information of the load, and recording the collected operation information;
inputting the collected operation information into a load characteristic generation module, generating VI track image characteristics and expert electrical characteristics, and forming and dividing an electricity load characteristic data set;
and classifying and correcting the collected operation information to generate an electricity probability model, and carrying out probability calculation to obtain a final load identification result.
S1: collecting operation information of the load, and recording the collected operation information;
further, the operation information comprises voltage and current signals of known loads, and records the time period of signal acquisition, the position of a sampling terminal and the operation information of the duration of load operation;
the voltage and current signals of the load comprise load steady-state voltage and current data with sampling frequency larger than 6.4kHz and sampling period larger than 10 power frequency periods.
It should be noted that, the sampling frequency is greater than 6.4kHz to reduce the distortion rate of the sampled current waveform, so that the sampled waveform fully contains the load information, and the sampling period is greater than 10 power frequency periods to fully obtain the steady-state waveform data of the load, so that there is enough load data for the recognition model to learn.
It should be noted that the current waveform of the household appliance load generally contains rich high frequency components such as high frequency pulses generated by the switching power supply, and the like. If the sampling frequency is too low, the high frequency components cannot be captured, which can cause serious distortion of the sampled current waveform, loss of a large amount of load information, influence the recognition accuracy, and generally, the sampling theorem requires that the sampling frequency is at least 2 times of the highest frequency in the signal. The main high-frequency component of the switch power supply waveform of the household appliance is above 3kHz, so that 6.4kHz is selected as the lowest sampling frequency, the sampling waveform can be well ensured not to be distorted, and richer load characteristic information is acquired.
It should be noted that the voltage and current waveforms of the household appliance will change greatly at the moment of starting, and then will tend to stabilize. If the sampling period is too short, the waveform in the transient process is likely to be captured, and the steady-state waveform is likely to be missed, so that the representativeness of the load characteristic learned by the recognition model is reduced, 10 power frequency periods (1/50 Hz=20ms) are selected as the minimum sampling period, the acquisition of the typical waveform under the steady state of the operation of the electrical appliance can be ensured, and the load characteristic information contained in the steady-state waveforms is rich, so that the recognition precision is improved. In conclusion, the high sampling frequency can acquire richer load high-frequency characteristics, the long sampling period can acquire more representative steady-state load characteristics, and the combination of the high sampling frequency and the long sampling period can acquire sampling waveforms containing rich load information, so that favorable conditions are provided for improving the load identification precision.
S2: inputting the collected operation information into a load characteristic generation module, generating VI track image characteristics and expert electrical characteristics, and forming and dividing an electricity load characteristic data set;
further, the VI track image features include image features formed by load voltage and current tracks, and the construction method is expressed as follows:
normalizing the load voltage and current waveforms:
wherein U is max 、I max Is the maximum value of voltage and current in one cycle, U min 、I min Is the minimum value of voltage and current in one cycle;
constructing a k x k zero matrix M with resolution k, and normalizing the voltageCurrent->And multiplying the resolution k, rounding downwards, taking the current value as a row, taking the voltage value as a column, and filling the corresponding position of the matrix M.
Further, the expert electrical characteristics comprise manual characteristics calculated according to load voltage and current data;
the manual characteristics comprise current peak value, current effective value, active power value, reactive power value, current harmonic distortion rate, current waveform coefficient, amplitude and phase of 1 st, 3 rd and 5 th current harmonic;
the current harmonic information is calculated by FFT, and the current peak value I is pp Calculation, expressed as:
I pp ＝max{i}-min{i}
max { i } is the maximum value in the current cycle of the sample, and min { i } is the minimum value in the current cycle of the sample;
effective value of current I rms Calculation, expressed as:
wherein i is n An instantaneous value of an nth sampling point in the current waveform;
active power P is calculated, expressed as:
wherein v is n An instantaneous value of an nth sampling point in the voltage waveform;
active power S is calculated, expressed as:
reactive power Q calculation, expressed as:
wherein u is n An instantaneous value of an nth sampling point in the voltage waveform;
power factor PF calculation, expressed as:
current harmonic distortion rate I THD Calculation, expressed as:
current waveform coefficient I wave Calculation, expressed as:
wherein N is the sampling frequency in one power frequency period.
Further, the dividing the power consumption load characteristic data set comprises dividing the generated power consumption load characteristic data set into a training set and a verification set, wherein the training and adjusting characteristic automatic enhancement and recognition module parameters, and stopping model training when the accuracy of the model in the verification set reaches a threshold value;
the feature enhancement network and the load identification network adopt an integral training mode, and a cross entropy loss function is adopted as a loss function of the feature enhancement network and the load identification network to train a model, and the model is expressed as follows:
wherein L is the total number of samples, M is the number of sample categories, y ic As a sign function, 1 is taken if the predicted class of the sample i is the same as the true class, 0 is taken if the predicted class of the sample i is different from the true class, and p is ic The probability that sample i belongs to category c.
S3: and classifying and correcting the collected operation information to generate an electricity probability model, and carrying out probability calculation to obtain a final load identification result.
Further, the electricity probability model comprises the following specific construction steps,
classifying known loads according to types, dividing a day into 24 time periods according to hours, counting the power consumption probability of each type of load in different time periods, and simulating a power consumption time period probability curve by using a Gaussian mixture model to obtain a power consumption time period model of the type c load, wherein the power consumption time period model is expressed as follows:
wherein n is the number of Gaussian models and the parameter theta 1 Comprises mean mu 1 ,…,μ n And standard deviation sigma 1 ,…,σ n ；
Fitting an electricity consumption time length probability curve of each type of load through a Gaussian mixture model, and representing the electricity consumption time length probability model of the type c load as phi 2c (x|θ 2 )；
For a home environment, the user potential is divided into: the living room, bedroom, dining room and bathroom 3 parts take the frequency of various types of loads in each environment as probability to obtain probability models of various types of loads in different positions, and the potential probability model of the type c load is expressed as phi 3c (x|θ 3 )；
For the load x identified as the type c, according to the collected electricity consumption period, electricity consumption duration and electricity consumption position, calculating to obtain the probability that the load x belongs to the type c, wherein the probability is expressed as follows:
P(x∈c)＝φ 1c (x|θ 1 )·φ 2c (x|θ 2 )·φ 3c (x|θ 3 )
wherein phi is 1c Gaussian probability model, θ, of the period of power consumption for type c loads 1 Is phi 1c Model parameters phi of (2) 2c Gaussian probability model, θ, of the power duration mixture for type c loads 2 Is phi 2c Model parameters phi of (2) 3c Potential probability model for type c load, θ 3 Is phi 3c Is used for the model parameters of the model.
Normalizing the probability of load generation to be distinguished and the probability of load generation of the known same type together to obtain the probability of each type output by the final classification correction module, wherein the probability is expressed as follows:
wherein P' represents the normalized probability value of the load to be discriminated, P represents the original probability value of the load to be discriminated, P max Representing the maximum of all the same type of load generation probabilities.
Further, the Gaussian mixture model comprises a fitting method which is a maximum expected algorithm, and the number k of the Gaussian mixture model is determined by setting a selection interval [ k ] of k min ,k max ]Sequentially select k min To k max Fitting is carried out on the number of Gaussian models, the number of Gaussian models when the correlation coefficient r of the fitting result is maximum is selected as the number k of the selected Gaussian models, and a calculation formula of the correlation coefficient r is expressed as follows:
wherein n is the length of the sequence, x is the actual load information data distribution, y is the fit probability curve of the Gaussian mixture model, x is the mean value of the actual load information data distribution, and y is the mean value of the fit probability curve of the Gaussian mixture model.
Example 2
A second embodiment of the application, which differs from the previous embodiment, is:
the functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present application may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or combination of the following techniques, as is well known in the art: discrete logic circuits having logic gates for implementing logic functions on data signals, application specific integrated circuits having suitable combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Example 3
Referring to fig. 3, a third embodiment of the present application provides a home load identification system based on feature automatic enhancement, which includes a load information acquisition module, a load feature generation module, a feature automatic enhancement and identification module, and a classification correction module;
the load information acquisition module is used for acquiring the running information of household loads;
the load characteristic generation module is used for generating VI track image characteristics and expert electrical characteristics to form an electricity load characteristic data set;
the characteristic automatic enhancement and recognition module is used for fusing the manually extracted electric characteristics and VI track image characteristics to obtain characteristics subjected to characteristic enhancement, and inputting the characteristics into the classification network to realize automatic recognition of different load types;
the classification correction module establishes power consumption probability models of different types of loads by adopting signal acquisition time period information, running position information and running time length information of the existing loads, and further carries out probability calculation on the judging results of the automatic characteristic enhancement and the identification module according to the information of the loads to be identified.
Example 4
In order to verify the beneficial effects of the application, scientific demonstration is carried out through economic benefit calculation and experiments.
The following parts are one embodiment of the present application, which provides a home load identification method based on feature automatic enhancement, comprising:
s1, respectively acquiring 100 groups of voltage and current waveforms of 11 household appliances such as a hot water kettle, an electric hair drier, a notebook computer and the like at a sampling frequency of 10kHz by using a load information acquisition module, and recording operation information such as a signal acquisition period, the position of a sampling terminal, the load operation duration and the like;
s2, inputting the collected load voltage and current signals into a load characteristic generating module, normalizing the voltage and the current to generate VI track image characteristics with the resolution of 32 multiplied by 32, calculating and generating expert electrical characteristics, and forming an electricity load characteristic data set with the sample number of 1100;
s3, dividing the generated power load characteristic data set into a training set and a verification set according to a ratio of 4:1, and using the training set and the verification set for training and adjusting characteristic automatic enhancement and identification module parameters. The characteristic enhancement network consists of 4 full-connection layers, one attention layer, 1 convolution layer and 1 deconvolution layer; the load recognition network is a LeNet-5 network, and the overall structure of the model is shown in figure 3. In the figure, the values in brackets of the full connection layer are represented as an input characteristic length and an output characteristic length, the values in brackets of the convolution layer are represented as a convolution kernel size and a convolution step size, the values in brackets of the deconvolution layer are represented as a deconvolution kernel size and a deconvolution step size, the values in brackets of the pooling layer are represented as a pooling kernel size and a pooling step size, and the values in brackets of the dropout layer are represented as the probability that each neuron connection is rejected. And setting the batch_size to be 32 in model training, and adopting an RMSProp optimizer to minimize a loss function based on a back propagation algorithm so as to realize iterative updating of model parameters. When the accuracy of the model in the verification set reaches 99%, model training is stopped;
s4, inputting the operation information of the known load into a classification correction module to obtain the probability that the load belongs to each type. When the calculated probability that the load belongs to the discrimination type of the characteristic automatic enhancement and recognition module is smaller than 0.5, multiplying the probability distribution output by the classification correction module by the probability distribution output by the characteristic automatic enhancement and recognition module, and selecting the type with the maximum probability as the final result of load recognition;
s5, load identification is achieved through each module of the system. The load to be identified is sequentially passed through a load information acquisition module, a load characteristic generation module, a characteristic automatic enhancement and identification module and a classification correction module, so that the identification of the load type can be realized.
The application uses 3 methods of original VI track characteristics, leNet-5 network, expert electrical characteristics, support vector machine and VI track, alexnet network as comparison groups to verify the identification effect of the algorithm, and the specific results are shown in Table 1.
TABLE 1
As can be seen from experimental results, the characteristic automatic enhancement algorithm of the application effectively enhances the characterization capability of VI track characteristics on load types, thereby effectively improving the effect of load identification, and has higher identification accuracy compared with other baseline models. Meanwhile, in different electric field scenes, the model of the application has the best recognition precision compared with other methods, and the application proves that the method can ensure the recognition performance in different and complex electric environments and has good robustness.
The application provides a load identification system and method based on characteristic automatic enhancement, which aims at solving the problems of low identification precision and poor method robustness in the existing household load identification field, enhances load characteristics by designing a characteristic self-adaptive enhancement network, and constructs a classification correction module to identify and correct loads with running rules contrary to identification types.
It should be noted that the above embodiments are only for illustrating the technical solution of the present application and not for limiting the same, and although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical solution of the present application, which is intended to be covered by the scope of the claims of the present application.
Claims (10)
1. A household load identification method based on characteristic automatic enhancement is characterized in that: comprising the steps of (a) a step of,
collecting operation information of the load, and recording the collected operation information;
inputting the collected operation information into a load characteristic generation module, generating VI track image characteristics and expert electrical characteristics, and forming and dividing an electricity load characteristic data set;
and classifying and correcting the collected operation information to generate an electricity probability model, and carrying out probability calculation to obtain a final load identification result.
2. A feature-based automatic enhanced home load identification method as claimed in claim 1, wherein: the operation information comprises voltage and current signals of a known load, and records the time period of signal acquisition, the position of a sampling terminal and the operation information of the duration of load operation.
3. A feature-based automatic enhanced home load identification method as claimed in claim 2, wherein: the VI track image features comprise image features formed by load voltage and current tracks, and the construction method comprises the following steps:
normalizing the load voltage and current waveforms:
wherein U is max 、I max Is the maximum value of voltage and current in one cycle, U min 、I min Is the minimum value of voltage and current in one cycle;
constructing a k x k zero matrix M with resolution k, and normalizing the voltageCurrent->And multiplying the resolution k, rounding downwards, taking the current value as a row, taking the voltage value as a column, and filling the corresponding position of the matrix M.
4. A feature-based automatic enhanced home load identification method as claimed in claim 3, wherein: the expert electrical characteristics comprise manual characteristics calculated according to load voltage and current data;
the manual characteristics comprise current peak value, current effective value, active power value, reactive power value, current harmonic distortion rate, current waveform coefficient, amplitude and phase of 1 st, 3 rd and 5 th current harmonic;
the current harmonic information is calculated by FFT, and the current peak value I is pp Calculation, expressed as:
I pp ＝max{i}-min{i}
max { i } is the maximum value in the current cycle of the sample, and min { i } is the minimum value in the current cycle of the sample;
effective value of current I rms Calculation, expressed as:
wherein i is n An instantaneous value of an nth sampling point in the current waveform;
active power P is calculated, expressed as:
wherein v is n An instantaneous value of an nth sampling point in the voltage waveform;
active power S is calculated, expressed as:
reactive power Q calculation, expressed as:
wherein u is n An instantaneous value of an nth sampling point in the voltage waveform;
power factor PF calculation, expressed as:
current harmonic distortion rate I THD Calculation, expressed as:
current waveform coefficient I wave Calculation, expressed as:
wherein N is the sampling frequency in one power frequency period.
5. A feature-based automatic enhanced home load identification method as claimed in claim 4, wherein: dividing the power consumption load characteristic data set comprises dividing the generated power consumption load characteristic data set into a training set and a verification set, automatically enhancing and identifying module parameters by training and adjusting characteristics, and stopping model training when the accuracy of the model in the verification set reaches a threshold value;
the feature enhancement network and the load identification network adopt an integral training mode, and a cross entropy loss function is adopted as a loss function of the feature enhancement network and the load identification network to train a model, and the model is expressed as follows:
wherein L is the total number of samples, M is the number of sample categories, y ic As a sign function, 1 is taken if the predicted class of the sample i is the same as the true class, 0 is taken if the predicted class of the sample i is different from the true class, and p is ic The probability that sample i belongs to category c.
6. A feature-based automatic enhanced home load identification method as claimed in claim 5, wherein: the electricity probability model comprises the following specific construction steps,
classifying known loads according to types, dividing a day into 24 time periods according to hours, counting the power consumption probability of each type of load in different time periods, and simulating a power consumption time period probability curve by using a Gaussian mixture model to obtain a power consumption time period model of the type c load, wherein the power consumption time period model is expressed as follows:
wherein n is the number of Gaussian models and the parameter theta 1 Comprises mean mu 1 ,…,μ n And standard deviation sigma 1 ,…,σ n ；
Fitting an electricity consumption time length probability curve of each type of load through a Gaussian mixture model, and representing the electricity consumption time length probability model of the type c load as phi 2c (x|θ 2 )；
For a home environment, dividing the potential use model into living rooms, bedrooms, restaurants and bathrooms, regarding the frequency of various types of loads in each environment as probability, obtaining probability models of the various types of loads in different positions, and representing the potential use probability model of the type c load as phi 3c (x|θ 3 )；
For the load x identified as the type c, according to the collected electricity consumption period, electricity consumption duration and electricity consumption position, calculating to obtain the probability that the load x belongs to the type c, wherein the probability is expressed as follows:
P(x∈c)＝φ 1c (x|θ 1 )·φ 2c (x|θ 2 )·φ 3c (x|θ 3 )
wherein phi is 1c Gaussian probability model, θ, of the period of power consumption for type c loads 1 Is phi 1c Model parameters phi of (2) 2c Gaussian probability model, θ, of the power duration mixture for type c loads 2 Is phi 2c Model parameters phi of (2) 3c Potential probability model for type c load, θ 3 Is phi 3c Model parameters of (2);
normalizing the probability of load generation to be distinguished and the probability of load generation of the known same type together to obtain the probability of each type output by the final classification correction module, wherein the probability is expressed as follows:
wherein P' represents the normalized probability value of the load to be discriminated, P represents the original probability value of the load to be discriminated, P max Representing the maximum of all the same type of load generation probabilities.
7. A feature-based automatic enhanced home load identification method as claimed in claim 6, wherein: the Gaussian mixture model comprises a maximum expected algorithm as a fitting method, and the number k of the Gaussian mixture model is determined by setting a selection interval [ k ] of k min ,k max ]Sequentially select k min To k max Fitting is carried out on the number of Gaussian models, the number of Gaussian models when the correlation coefficient r of the fitting result is maximum is selected as the number k of the selected Gaussian models, and a calculation formula of the correlation coefficient r is expressed as follows:
wherein n is the length of the sequence, x is the actual load information data distribution, y is the fit probability curve of the Gaussian mixture model,for the mean value of the actual load information data distribution, +.>And fitting the mean value of the probability curve to the Gaussian mixture model.
8. A system employing a feature-based automatic enhanced home load identification method as claimed in any one of claims 1 to 7, characterized in that: the system comprises a load information acquisition module, a load characteristic generation module, a characteristic automatic enhancement and identification module and a classification correction module;
the load information acquisition module is used for acquiring the running information of household loads;
the load characteristic generation module is used for generating VI track image characteristics and expert electrical characteristics to form an electricity load characteristic data set;
the characteristic automatic enhancement and recognition module is used for fusing the manually extracted electric characteristics and VI track image characteristics to obtain characteristics subjected to characteristic enhancement, and inputting the characteristics into the classification network to realize automatic recognition of different load types;
the classification correction module establishes power consumption probability models of different types of loads by adopting signal acquisition time period information, running position information and running time length information of the existing loads, and further carries out probability calculation on the judging results of the automatic characteristic enhancement and the identification module according to the information of the loads to be identified.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any of claims 1 to 7 when the computer program is executed.
10. A computer-readable storage medium having stored thereon a computer program, characterized by: the computer program implementing the steps of the method of any of claims 1 to 7 when executed by a processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310759240.9A CN117039840A (en) | 2023-06-26 | 2023-06-26 | Feature automatic enhancement-based household load identification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310759240.9A CN117039840A (en) | 2023-06-26 | 2023-06-26 | Feature automatic enhancement-based household load identification method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117039840A true CN117039840A (en) | 2023-11-10 |
Family
ID=88634283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310759240.9A Pending CN117039840A (en) | 2023-06-26 | 2023-06-26 | Feature automatic enhancement-based household load identification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117039840A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117333724A (en) * | 2023-11-28 | 2024-01-02 | 天津滨电电力工程有限公司 | Non-invasive load identification method based on multi-feature fusion image |
-
2023
- 2023-06-26 CN CN202310759240.9A patent/CN117039840A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117333724A (en) * | 2023-11-28 | 2024-01-02 | 天津滨电电力工程有限公司 | Non-invasive load identification method based on multi-feature fusion image |
| CN117333724B (en) * | 2023-11-28 | 2024-02-27 | 天津滨电电力工程有限公司 | Non-invasive load identification method based on multi-feature fusion image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Angelis et al. | NILM applications: Literature review of learning approaches, recent developments and challenges | |
| Sirojan et al. | Deep neural network based energy disaggregation | |
| CN110514889A (en) | A kind of method and system of non-intrusion type household electricity remained capacity | |
| CN109359665B (en) | Household appliance load identification method and device based on support vector machine | |
| CN109345409A (en) | A kind of resident's comprehensive energy efficiency management method based on bandwidth carrier | |
| CN110416995B (en) | Non-invasive load decomposition method and device | |
| CN111563827B (en) | Load decomposition method based on physical characteristics of electrical appliance and electricity utilization behaviors of residents | |
| CN117039840A (en) | Feature automatic enhancement-based household load identification method and system | |
| CN109470957B (en) | Non-invasive electrical appliance identification method based on Frey's distance | |
| TWI526852B (en) | Method for counting number of people based on appliance usages and monitoring system using the same | |
| Dash et al. | An appliance load disaggregation scheme using automatic state detection enabled enhanced integer programming | |
| Jiao et al. | A context-aware multi-event identification method for nonintrusive load monitoring | |
| CN113807225A (en) | Load identification method based on feature fusion | |
| Anderson | Non-intrusive load monitoring: Disaggregation of energy by unsupervised power consumption clustering | |
| CN114444539A (en) | Power load identification method, apparatus, device, medium, and program product | |
| CN112085111A (en) | Load identification method and device | |
| TWI492182B (en) | A method for identifying electric appliance loads | |
| CN117172589A (en) | Industry typical user power demand response potential evaluation method | |
| CN112614004A (en) | Method and device for processing power utilization information | |
| CN115983347A (en) | Non-invasive load decomposition method, device and storage medium | |
| CN111898694A (en) | Non-invasive load identification method and device based on random tree classification | |
| CN115932435A (en) | Resident non-invasive load monitoring method based on low-frequency acquisition signals | |
| CN116561569A (en) | Industrial power load identification method based on EO feature selection and AdaBoost algorithm | |
| CN114355275A (en) | Electric energy meter load monitoring method, system and device and computer readable storage medium | |
| Li et al. | A training-free non-intrusive load monitoring approach for high-frequency measurements based on graph signal processing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |